Random-Shape Block Prediction With Mixed Intra-Inter Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression technologies face challenges in efficiently encoding and decoding video data due to increasing data volumes and the need for improved encoding efficiency and higher image quality, particularly with non-rectangular blocks.
Innovation Solution
The method involves partitioning current blocks into non-rectangular blocks for inter and intra prediction, determining block types based on syntax elements, and applying adaptive filtering to improve encoding and decoding efficiency, including weight-based prediction value integration and simplified filtering processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If video data is compressed using existing technologies (H.264/AVC, HEVC), then encoding efficiency is improved, but data volume continues to increase due to higher resolution and frame rate requirements
Solution Approach 1:
The current block is partitioned into multiple non-rectangular blocks (first non-rectangular block and second non-rectangular block) with different shapes and orientations. This segmentation allows each partition to be processed independently with optimized prediction modes, improving compression efficiency while handling high-resolution video data
Solution Approach 2:
Different prediction modes are applied to different non-rectangular blocks based on their local characteristics. The first non-rectangular block uses one prediction mode while the second uses another, allowing local optimization of compression quality according to the specific content and directionality of each block region
2Ease of operation
If only inter prediction is performed on non-rectangular blocks, then encoding simplicity is maintained, but prediction accuracy and image quality deteriorate
Solution Approach 1:
The system dynamically selects between inter prediction and intra prediction modes for each non-rectangular block based on its characteristics. The prediction mode is not fixed but adapts to the local content requirements, improving accuracy while maintaining reasonable encoding complexity through conditional mode selection
3Manufacturing precision
If intra prediction is performed on non-rectangular blocks with reference to inter prediction values, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent combines inter prediction and intra prediction processes by allowing intra prediction to reference inter prediction values. This merging of prediction methods leverages the motion compensation strength of inter prediction while adding the local adaptability of intra prediction, achieving higher accuracy without requiring completely separate processing pipelines
4Manufacturing precision
If adaptive filtering is applied to remove block edge discontinuity, then image quality is improved, but processing time and complexity increase
Solution Approach 1:
Adaptive filtering is applied selectively rather than uniformly across all blocks. The filtering operation is performed partially on blocks where discontinuity is detected, using weight-based blending of prediction values at block edges. This partial application reduces processing time while still improving image quality where needed
Data Source
AI summary
A method and a device for inversely transforming transform coefficients of a current block are disclosed. A method for predicting a current block according to a first mode, includes the steps of: on the basis of a partition mode syntax element, partitioning the current block into non-rectangular blocks; determining an intra-predicted intra block and an inter-predicted inter block among the non-rectangular blocks; and deriving prediction samples of a first area including the inter block on the basis of motion information, and deriving prediction samples of a second area including the intra block on the basis of an intra prediction mode.